EUR 23031 EN - 2007

Certification of the Deuterium-to-Hydrogen (D/H) ratio in a 1,1,3,3-tetramethylurea master batch

IRMM-425R. Zeleny, H. Emteborg, F. Ulberth

The mission of the IRMM is to promote a common and reliable European measurement system in support of EU policies. European Commission Joint Research Centre Institute for Reference Materials and Measurements Contact information Address: R. Zeleny E-mail: Reinhard.zeleny@ec.europa.eu Tel.: 014/571 615 Fax: 014/571 548 http://www.irmm.jrc.be/html/homepage.htm http://www.jrc.ec.europa.eu Legal Notice Neither the European Commission nor any person acting on behalf of the Commission is responsible for the use which might be made of this publication. A great deal of additional information on the European Union is available on the Internet. It can be accessed through the Europa server http://europa.eu/ EUR 23031 EN ISBN 978-92-79-07550-6 ISSN 1018-5593 DOI 10.2787/52761 Luxembourg: Office for Official Publications of the European Communities © European Communities, 2007 Reproduction is authorised provided the source is acknowledged Printed in Belgium

Certification of the Deuterium-to-Hydrogen (D/H) ratio in a 1,1,3,3-tetramethylurea master batch

IRMM-425

R. Zeleny, H. Emteborg, F. Ulberth

European Commission, Joint Research Centre Institute for Reference Materials and Measurements (IRMM)

2440 Geel, Belgium

1

Summary This report describes the production of a tetramethylurea reference material (IRMM-425),

certified for its deuterium-to-hydrogen (D/H) ratio. The material is to be used as an internal

standard in site-specific natural isotope fractionation – nuclear magnetic resonance (SNIF-

NMR) spectroscopy measurements for determining the D/H ratios of ethanol distilled from

wines, an important measure in wine authenticity testing (Commission Regulation 2676/90,

[1]).

Commercially obtained TMU with a sufficiently high D/H ratio (>120 x 10-6) was purified by

removing most of the residual water and filled into amber glass ampoules in 7.8 mL portions.

Homogeneity was tested and no heterogeneity observed. Stability studies indicated no

material degradation for 4 weeks at 60 ºC and for 18 months at –20 ºC. Batch

characterisation was accomplished in an inter-laboratory comparison using the SNIF-NMR

technique exclusively.

The certified value was obtained as the unweighted mean of the laboratory means of the

accepted sets of results. The expanded uncertainty associated (k = 2) includes contributions

from (potential) heterogeneity, potential instability, characterisation, as well as the carried-

over uncertainty from BCR-656 (96% wine ethanol) that was used as the internal standard in

the measurements. Due to its negligible contribution, the uncertainty associated with the

purity of the material did not need to be included.

The certified value and its associated uncertainty are:

Amount-of-substance fraction

Certified value 1,2,4) Uncertainty 3)

Deuterium-to-hydrogen (D/H) ratio 141.9 x 10-6 0.7 x 10-6 1) Traceable to V-SMOV (Vienna Standard Mean Ocean Water) by strictly adhering to the Community reference

method (SNIF-NMR, site-specific natural isotope fractionation – nuclear magnetic resonance spectroscopy) as defined in Annex 8, Commission Regulation 2676/90. For V-SMOW, a value of 155.76 ppm was used (Tellus (1970) 22: 712-715).

2) This value was obtained by an inter-laboratory comparison (11 laboratories) employing SNIF-NMR at 46.1 (300), 61.4 (400), and 76.7 (500) MHz for 2H (1H). The value has been corrected for the material impurities (including water).

3) Expanded uncertainty with a coverage factor of k = 2, according to the Guide to the Expression of Uncertainty in Measurement, corresponding to a level of confidence of about 95 %.

4) The certified value has a measurement unit of one; it is commonly expressed in parts per million (ppm).

This CRM constitutes a master batch and will be used for the certification of secondary

batches; the material is not for sale.

2

3

Table of Contents Summary 1 Table of contents 3 Glossary 4 1 Introduction 5 2 Time table of the project 9 3 Participants 10 4 Processing 11 5 Methodology used 13 6 Homogeneity 16 6.1 Homogeneity 16

6.2 Minimum sample intake 17

7 Stability studies 18 7.1 Short-term stability 18

7.2 Long-term stability 19

8 Characterisation 21 8.1 Discussion of results 21

8.2 Certified value and uncertainty budget 24

9 Metrological traceability 26 10 Instructions for use 26 10.1 Intended use 26

10.2 Storage 26

10.3 Minimum sample intake 26

10.4 Use of the certified values 26

10.5 Safety precautions 26

11 Acknowledgements 27 12 Annexes 27 13 References 28 Annex A – homogeneity data 30 Annex B – short-term stability data 32 Annex C – long-term stability data 34 Annex D – characterisation data 36 Annex E – characterisation – method-related information 37

4

5

Glossary ANOVA Analysis Of Variance b Slope of the linear regression BCR Community Bureau of Reference BEVABS European Office for Wine, Alcohol and Spirit Drinks CAP Certification Advisory Panel CI Confidence interval CRM Certified reference material D Deuterium (2H) D/H Deuterium-to-hydrogen ratio δ Chemical shift df Degrees of freedom EA-P Elemental analysis – pyrolysis EU European Union FID Flame ionization detection GC Gas chromatography IHCP Institute for Health and Consumer Protection IRMM Institute for Reference Materials and Measurements IRMS Isotope ratio mass spectrometry ISO International Organisation for Standardization JRC Joint Research Centre k Coverage factor KFT Karl Fischer titration m/m Mass/mass MHz Mega hertz MS Mass spectrometry MSbetween Mean of squares between groups (ANOVA) MSwithin Mean of squares within groups (ANOVA) NIST National Institute of Standards and Technology ppm Parts per million QC Quality Control RSD Relative standard deviation s Standard deviation sbb Between-bottle (in)homogeneity standard deviation SNIF-NMR Site-specific natural isotope fractionation – nuclear magnetic

resonance swb Within-bottle standard deviation t t value

Dmt Alcoholic grade of BCR-656, expressed in mass %

TMU Tetramethylurea u*bb Standard uncertainty due to the inhomogeneity that can be hidden by

the method repeatability uBCR-656a Uncertainty of D/HI value of BCR-656 uBCR-656b Uncertainty of the D

mt value of BCR-656 uchar Standard uncertainty from characterisation UCRM Expanded uncertainty associated to the CRM ults Standard uncertainty associated to long-term (in)stability V-SMOW Vienna Sea Mean Ocean Water

6

1 Introduction

The isotopic ratio of deuterium to hydrogen (D/H) is regularly measured in ethanol distilled

from wine to detect wine adulterations, such as mixing high with low quality wines, or the

addition of sugar before fermentation (chaptalisation) outside the allowed limits. The

Community reference method to determine the wine ethanol D/H ratio, specified in

Commission Regulation 2676/90 [1], is Site-specific Natural Isotope Fractionation measured

by Nuclear Magnetic Resonance (SNIF-NMR®) [2]. To monitor wine authenticity, the

measured values are compared with those from the respective authentic wines listed in the

European Union (EU) Wine Databank [3], operated by the European Office for Wine, Alcohol

and Spirit Drinks (BEVABS) at the Joint Research Centre (JRC) of the European

Commission.

The reference method specifies the internal standard in the SNIF-NMR measurements to be

tetramethylurea (TMU, figure 1) with a known D/H ratio, available as a certified reference

material (CRM) from the Institute for Reference Materials and Measurements (IRMM) –

current batch "STA-003". This compound possesses appropriate physico-chemical

properties, such as a low volatility, and excellent miscibility with wine ethanol. Furthermore, it

exhibits a convenient chemical shift in NMR analysis to avoid interference with ethanol peaks

and it yields a signal width similar to that of the ethanol methyl signal.

Figure 1. 1,1,3,3 – Tetramethylurea, CAS Number: 632-33-4

The first TMU reference material was certified measuring mixtures of TMU with V-SMOW

(Vienna - Standard Mean Ocean Water) by SNIF-NMR [4]; all subsequent TMU batches,

however, were certified using reference ethanols, which themselves were calibrated against

the penultimate TMU batch. This procedure has some drawbacks, such as the dependence

of the certified value on a former material, resulting in an increasing uncertainty and an

elongated traceability chain from batch to batch. In addition, value assignment was

accomplished using only three dedicated NMR spectrometers, which is reflected in the high

coverage factor (4.3) of the expanded uncertainty.

7

IRMM strived for revising the certification strategy by producing a TMU master batch, to

which all future secondary batches shall be linked. An independent property value

assignment was envisaged for shortening the traceability chain and lowering the uncertainty

of the certified value. Isotope Ratio Mass Spectrometry (IRMS) was intended to be used,

which allows a direct link to internationally recognised standards such as V-SMOW, and

presumably reveals results with a lower overall uncertainty due to highly precise

measurements obtainable by IRMS.

Prior to application of this concept, key prerequisites had to be verified and confirmed: IRMS

results need to be in agreement with SNIF-NMR results, because value consistency is

indispensable to maintain validity of the wine databank. Furthermore, the between-lab

variation has to be reasonably small (in the range of that from SNIF-NMR data) to keep the

uncertainty of the certified value at an acceptable level. Therefore, an inter-comparison was

conducted to evaluate the currently obtainable method performance of the EA-P-IRMS

(Elemental Analysis–Pyrolysis–IRMS) methodology and to evaluate its applicability for

characterisation measurements of TMU. The results reveal that expert laboratories are able

to produce highly precise and comparable data, which demonstrates the potential of IRMS in

food authenticity control. However, a systematic bias between IRMS and NMR results was

observed, which rules out the IRMS technique for characterisation measurements of the

TMU master batch [5]. Although further measurements are needed to identify possible

reasons for this finding, it seems that the observed bias cannot be attributed solely to

differences in the measurement principles, but is also compound-dependent [6].

The finally chosen strategy for the certification of the TMU master batch is the following:

TMU-ethanol mixtures are analysed by SNIF-NMR. As the ethanol, BCR-656 with a certified

D/HI and D/H II value is used [7]. Taking the peak heights and the masses of TMU and

ethanol into account, the D/H ratio of TMU can then be calculated using the formula

stipulated in Commission Regulation 2676/90 [1] and depicted in chapter 5. The originally

envisaged independence of the certified value from previous CRMs is thus not achieved, and

the uncertainty of the D/HI value of BCR-656 has to be taken up in the uncertainty budget;

however, this was deemed a necessary and suitable compromise due to the unavailability of

IRMS for this application, and the fact that not data accuracy (true value of the D/H ratio), but

data consistency to previous TMU batches has to be guaranteed to maintain and safeguard

the validity of the European Wine Databank. Furthermore, the master batch represents a

new anchor point, so neither the length of the traceability chain nor the uncertainty will

increase for future secondary batches. Finally, a considerably larger number of laboratories

(11) were foreseen in this certification campaign for performing the characterisation

measurements, which substantially reduces the coverage factor and thus the overall

8

uncertainty. It shall be noted that the certified value and its uncertainty are expressed as a

ratio (141.9 ± 0.7 x 10-6) with the measurement unit of one; commonly SNIF-NMR results are

expressed in parts per million (ppm).

9

2 Time table of the project

Activity Finalised by Procurement of base material May 2005 Processing September 2005 Homogeneity testing January 2006 Short-term stability testing January 2006 Long-term stability testing March 2007 Characterisation April 2007

10

3 Participants

Processing

Eurofins Scientific Analytics, Nantes FR European Commission, Joint Research Centre, Institute for Reference Materials and Measurements, Geel BE (ISO Guide 34 accreditation; BELAC 268-TEST)

Homogeneity and stability studies

Bundesinstitut für Risikobewertung, Berlin DE (ISO/IEC 17025 accreditation; AKS-P22001-EU)

Characterisation*

Bundesinstitut für Risikobewertung, Berlin DE (ISO/IEC 17025 accreditation; AKS-P22001-EU) Central Science Laboratory, York GB (ISO/IEC 17025 accreditation; UKAS 1642) Eurofins Scientific Analytics, Nantes FR (ISO/IEC 17025 accreditation; COFRAC Nº 1-0287) European Commission, Joint Research Centre, Institute for Health and Consumer Protection, Ispra IT (ISO/IEC 17025 accreditation, SINAL 0387) Instituto Nacional de Engenharia, Tecnologia e Inovação, Lisboa PT (ISO/IEC 17025 accreditation; IPAC L-0324) Laboratoire Interrégional des Douanes de Paris FR Landesuntersuchungsamt Rheinland-Pfalz, Speyer DE (ISO/IEC 17025 accreditation; SAL-RLP-L 19.04.03) Ministry of Health, State General Laboratory, Nicosia CY (ISO/IEC 17025 accreditation; ESYD 290) National Institute of Chemistry, Ljubljana SI *In alphabetical order

Project Management & Data evaluation

European Commission, Joint Research Centre, Institute for Reference Materials and Measurements, Geel BE

(ISO Guide 34 accreditation; BELAC 268-TEST)

11

4 Processing of the material

A batch of 100 litres of TMU was acquired in China with a sufficiently high deuterium-to-

hydrogen (D/H) ratio (> 120 ppm). This material, which contained about 0.5 % m/m of water,

was further processed at Eurofins Scientific Analytics to remove most of the water from the

product, as water initiates hydrolysis of TMU to trimethylurea and other degradation

products. Effective removal of most of the water was accomplished by adding molecular

sieve to the TMU.

The identity of TMU was investigated at Eurofins performing GC-MS measurements using an

ion-trap mass spectrometer, and comparing the obtained spectrum with the NIST library

spectrum of TMU. For all 3 independently sampled aliquots, identity could be confirmed.

Purity assessment was accomplished at Eurofins employing GC-FID, 1H-NMR, and 13C-

NMR. GC-FID gave an average value of 99.89 ± 0.01 % (n = 3), calculated as TMU area to

total peak area in the chromatogram. 1H NMR (n = 3) and 13C NMR (n = 3) agreed with

literature data and did not show any significant impurity.

The final product exhibited a water content of 0.08 ± 0.01 % m/m (n = 5) as determined by

Karl Fischer titration (KFT; shown in Table 1), resulting in a corresponding to a purity of

99.92 % m/m. The sum of impurities detected by GC-FID and KFT was calculated, and the

corresponding purity of TMU yielded 99.81 %. The uncertainty of the combined impurities

detected by GC-FID and KFT was calculated as the square root of the sum of the squares of

the 2 contributions, and yielded 0.01 %.

5 L of the purified TMU, aliquoted in 10 bottles (brown glass, screw cap, additional wax

stopper; see figure 2) were shipped to IRMM and stored at -20 ºC until ampouling. Three

small feasibility studies were carried out to find the best possible conditions of ampouling the

slightly volatile and hygroscopic TMU. Details can be found in [8].

Ampouling was finally accomplished as follows: 555 amber 10 mL ampoules were opened

and stored in desiccators one day prior to filling. The ten 500 mL TMU bottles (depicted in

Fig. 2) were taken out of the refrigerator the evening before filling to equilibrate at 20 °C in

the dark. A dispenser and a 10 L borosilicate bottle were set up for operation. The inlet

(suction-side) of the dispenser was constantly flushed with argon. The 10 L flask was first

rinsed with 300 mL of TMU from one of the Eurofins bottles, and another 100 mL of TMU

was used to rinse the dispenser. Thereafter, the content of the ten TMU bottles was pooled

12

in the 10 L flask. Finally, ampoules were filled with 7.6 g (7.8 mL) of TMU and transferred to

an ampouling machine for flame-sealing (ROTA R 910/PA, Wehr, Germany). The first 18

ampoules were discarded, and 531 ampoules were labelled (order of filling). Finally, the

samples were stored at -20 ºC

Figure 2. Dispenser and 10 L flask (left), TMU as provided by Eurofins (right)

Quality control after processing

After processing, five samples were picked using a random-stratified sample picking scheme

for water analysis by coulometric Karl Fischer titration. Table 1 shows the results, confirming

the expected low water levels in the final TMU candidate reference material.

Table 1. Water content in IRMM-425 as measured by Karl Fischer titration

Bottle n. Water content*,

[g/100 g]

67 0.078

122 0.078

242 0.077

363 0.078

509 0.079

* Mean of two replicates

13

5 Analytical methodology used

All measurements were accomplished by employing site-specific nuclear isotope

fractionation-nuclear magnetic resonance (SNIF-NMR) spectroscopy in accordance with

Commission Regulation 2676/90 [1]. BCR-656 (96 vol% ethanol) was provided by IRMM and

used as the internal standard. This material is certified for two site-specific ratios of

deuterium to hydrogen in the ethanol molecule: D/HI refers to the peak to the right of the

TMU peak (methyl group, CH2DCH2OH), whereas D/HII refers to the peak to the left of the

TMU peak (methylene group, CH3CHDOH), [1]. The OH-group signal (CH3CH2OD) was not

evaluated (figure 3).

Figure 3. Representative 2H spectrum of ethanol/TMU mixture. Ethanol reveals three peaks, reflecting the position of the deuterium in the molecule Measurements were acquired on 300, 400 or 500 - MHz NMR appliances. 10 mm diameter

NMR tubes were used.

For sample preparation, the TMU ampoules were opened and two NMR tubes were prepared

immediately per ampoule (duplicate analysis). Sample preparation was done by weighing in

CH3CH2OD

=(D/H)III

CH2DCH2OH

=(D/H)I

TMU

CH3CHDOH

=(D/H)II

δ (ppm)

Inte

nsity

(arb

itrar

y un

it)

CH3CH2OD

=(D/H)III

CH2DCH2OH

=(D/H)I

TMU

CH3CHDOH

=(D/H)II

δ (ppm)

Inte

nsity

(arb

itrar

y un

it)

14

ca. 3.2 mL of reference alcohol BCR-656 (ca. 2.5 g) and ca. 1.3 mL of TMU (ca. 1.3 g);

finally, some 100 - 200 μL (ca. 0.2 – 0.3 g) of the lock substance hexafluorobenzene was

added. All exact weights and sample numbers were recorded and reported. In addition, date

and time of measurements were documented.

SNIF-NMR instrument performance was verified by analysing laboratory QC samples; most

frequently, BCR-123 (3 different ethanols from beet, grape, and cane origin) was used [9];

some laboratories also used in-house ethanols distilled from wine with an assigned D/H

value.

One analysis consisted of 10 recorded spectra. Finally, data processing was accomplished

by Fourier transformation, line broadening and automated baseline correction. Peak heights

were determined by the respective software used (details are given in Annex D).

The formulae depicted below and stipulated in [1] are used to calculate the ethanol D/H

values using the certified TMU D/H value.

Dm

St

A

stII t

HDmmTHD )/(5866.1)/( ×××= D

m

St

A

stIIII t

HDmmTHD )/(3799.2)/( ×××=

with TI, height of signal I (CH2DCH2OH) divided by height of signal of TMU TII, height of signal II (CH3CHDOH) divided by height of signal of TMU mst, mass of internal standard (TMU) to nearest 0.1 mg mA, mass of distilled wine ethanol (BCR-656) to nearest 0.1 mg (D/H)St, certified value of TMU

Dmt , strength by mass of the wine ethanol in % (m/m), whereas

100'×

−=

pppt D

m , whereas

p, mass of wine ethanol p’, mass of water content in the wine ethanol determined by Karl Fischer titration.

The factors of 1.5866 and 2.3799, respectively, derive from:

I

St

St

A

PP

MM

×=5866.1 II

St

St

A

PP

MM

×=3799.2

with MA, molecular mass of ethanol MSt, molecular mass of internal standard (TMU) PSt, number of hydrogen atoms in TMU molecule PI, number of hydrogen atoms at site I of the ethanol molecule PII, number of hydrogen atoms at site II of the ethanol molecule

15

In this exercise, the formula was inversed to calculate the TMU D/H value using either the

certified D/HI or D/HII value of BCR-656.

The values and their corresponding uncertainties of BCR-656 [6] used in this study are listed

in Table 2:

Table 2. Relevant certified values and uncertainties of BCR-656

Certified value Uncertainty

D/HI in ppm 102.84 0.20 Dmt in % (m/m) 94.61 0.05

Preferentially, the D/HI value of BCR-656 is used for the calculation of the TMU value, due to

the favourable stoechiometric factors - 3 H in the case of D/HI (CH2D-CH2-OH) signal, to 12

H in TMU - compared to 2 H in the case of D/HII (CH3-CHD-OH) signal to 12 H in TMU. An

investigation comparing different methods of data evaluation revealed that the official method

[1] tends to slightly overestimate the D/HII ratio due to field distortion effects [10], which also

underpins our decision to use solely the certified value of D/HI of BCR-656 for calculating the

D/H ratio of the TMU master batch and the uncertainty contributions.

Consequently, only the values calculated using the D/HI of BCR-656 were further on taken

into account (homogeneity, stability, and characterisation measurements).

16

6 Homogeneity and minimum sample intake

6.1 Homogeneity Eight samples (about 1.5 % of the batch) were selected from the produced batch using a

random stratified sample picking scheme. Samples were analysed in duplicate in a

randomised order to allow detection of possible trends in the analytical sequence or the filling

order, using SNIF-NMR as described above (400 MHz instrument). Care was taken that not

more than 7 measurements were performed in between two replicate measurements of one

given sample. This was a technical requirement as the two replicates from one sample

(ampoule) are prepared at the same time, and due to the volatility of TMU evaporation might

occur in the non-sealed NMR tubes (C. Fauhl, Berlin, personal communication).

Measurements were performed under repeatability conditions.

Results were evaluated using SoftCRM [11] to detect possible trends regarding the filling

sequence or analytical sequence and to estimate the uncertainty contribution of potential

heterogeneity. Data were scrutinised for outliers and normal distribution using normal

probability plots. No outliers were detected, and values were found to be normally distributed.

The slopes of the regression lines were tested for their significance at 95 % and 99 % level of

confidence: no significant trend was observed for neither filling order nor analytical sequence

(Table 3).

Table 3. Linear regression and statistical parameters associated to the homogeneity evaluation of IRMM-425

Statistical parameters Filling sequence

Analytical sequence

Slope (b) -0.00027 -0.00621

Standard error of slope (sb) 0.000156 0.00787

Degrees of freedom (df) 6 14

|b|/sb 1.74 0.79

t(0.05;6) 2.45 -

t(0.05;14) - 2.14

Statistical significance (95% confidence interval) No No

Possible heterogeneity is hidden by the method repeatability (ANOVA; MSbetween < MSwithin),

which made it necessary to calculate u*bb to estimate the uncertainty contribution for possible

in-homogeneity using the following formula [12]:

17

4* 2

dfnRSDu method

bb =

RSDmethod represents the method repeatability in %, n the number of replicates per sample,

and df the degrees of freedom taken from the ANOVA results table. RSDmethod was calculated

using the following formula:

100×=y

MSRSD within

method

whereas MSwithin represents the mean of squares within groups from the ANOVA table, and

y the overall mean of the homogeneity measurements.

The value obtained was u*bb = 0.06 % using the certified D/HI value of BCR-656. The

calculated u*bb holds for the sample intake laid down in Commission Regulation 2676/90, that

is about 2.5 g of ethanol and about 1.3 g of TMU, using a 10 mm diameter NMR tube

(standard dimension).

Table 4. Summary of the homogeneity test

Average sbb sbb % swb swb % u*bb u*bb % D/H ratio 141.828 n.c. n.c. 0.169 0.119 0.084 0.060

n.c. cannot be calculated as MSbetween < MSwithin

The estimated value for u*bb is close to zero, as can be expected for a solution. It can thus be

concluded that IRMM-425 is a homogeneous material.

6.2 Minimum sample intake This parameter was not assessed, as the method predefines the sample intake (Commission

Regulation 2676/90): for 10 mm diameter NMR tubes, 1.3 g have to be used.

18

7 Stability studies

7.1 Short-term stability

A four weeks isochronous study [13] was performed to evaluate stability of the TMU master

batch during transport. Samples were selected from the produced batch using a random

stratified sample picking scheme.

Samples were stored at +18 °C and +60 °C and at a reference temperature of -20 °C. Three

ampoules were stored at each temperature for 0, 1, 2, and 4 weeks. After the indicated

storage periods, the samples were transferred to storage at -20 °C until analysis. Samples

were analysed in duplicate in the order predefined at IRMM (randomised sample order) using

SNIF-NMR as described above (400 MHz instrument). Again, care was taken that not more

than 7 measurements were performed in between two replicates of one given sample (see

chapter homogeneity). Measurements were performed under repeatability conditions.

SoftCRM [11] was used to evaluate the measurement results for possible significant trends

(degradation, enrichment) due to the storage conditions. One outlier was detected using the

Grubbs test at 18 ºC (95 % level of confidence), whereas no outlier was detected at 60 ºC.

The outlier was not excluded as data scrutiny revealed no technical reason to do so.

The measurement results and resulting graphs are presented in Annex B. The observed

slopes were tested for significance using a t-test, with tα,df being the critical t-value (two-

tailed) for a confidence level α = 0.05 (95 % confidence interval). The slope was considered

as statistically significant when b/sb > t α,df . No significant slope at 95 % level of confidence

was detected for both temperatures tested (Table 5).

Table 5. Linear regression and statistical parameters associated to the short term stability evaluation of IRMM-425

Storage Temperature 18 ºC 60 ºC

Statistical parameters

Slope (b) -0.015 0.016

Standard error of slope (sb) 0.038 0.048

Degrees of freedom (df) 22 22

|b|/sb 0.39 0.33

t(0.05;22) 2.07 2.07

Statistical significance (95% confidence interval) No No

19

Conclusion: no detectable degradation was observed, even at storage conditions of 60 °C for

4 weeks. Therefore no special precautions are required during transport; the material can be

dispatched under ambient conditions.

7.2 Long-term stability

Another isochronous study was performed to evaluate long-term stability of the TMU master

batch. Samples were selected from the produced batch using a random stratified sample

picking scheme.

Samples were stored at -20 °C and at a reference temperature of -70 °C. The reasoning was

that the final material is envisaged to be stored at -20 ºC, and therefore it was not evaluated

whether long-term stability can be proven also for higher temperatures.

6 ampoules were stored at each temperature for 0, 12, and 18 months. After the indicated

storage periods, the samples were transferred to storage at -70 °C until analysis. Samples

were analysed in duplicate in the order predefined at IRMM (randomised sample order) using

SNIF-NMR as described above (400 MHz instrument). Again, care was taken that not more

than 7 measurements were performed in between two replicates of one given sample (see

above). Measurements were performed under repeatability conditions.

Data evaluation using SoftCRM [11] was done in the same way as for the short-term stability

study. No outliers were detected. Table 6 summarizes the results.

Table 6. Evaluation of long-term stability study, 18 months

Storage Temperature -20 ºC

Statistical parameters

Slope (b) 0.0030

Std error slope (sb) 0.0043

df 34

|b|/sb 0.71

t(0.05;34) 2.03

Statistical significance (95 % confidence interval) No

Finally, the uncertainty contribution ults due to (in)stability was estimated for a shelf-life of 5

years and found to be 0.18 % (temperature: -20 ºC) using the following formula [14]:

xxx

RSDui

lts *)( 2∑ −

=

20

where

xi are the time points for each replicate

x is the shelf life (in months)

RSD is the relative standard deviation of all data in this series

The relationship between ults, and shelf-life is illustrated Annex C.

Conclusion:

The TMU master batch did not exhibit any significant instability at a temperature of -20 °C for

a period of 18 months. Stability data available from previous TMU batches support these

findings. Additional studies to assess stability over a longer period of time are ongoing.

Those results will be used for extension of the material's shelf life. In conclusion, the material

has shown to be stable when stored at -20 ºC.

21

8 Characterisation

Batch characterisation was accomplished in an inter-laboratory comparison. Laboratories

needed to have proven experience in the isotopic analysis of wine ethanol samples by SNIF-

NMR and were selected according to the quality criteria required by IRMM. In several cases,

laboratories were accredited to ISO 17025; otherwise, the required measurement

performance was demonstrated by successful participation in PT schemes, respective quality

control charts, and scientific publications in the field.

8.1 Discussion of results

Laboratories had to comply strictly with 2676/90 [1] and thus used the methodology outlined

above. Small method variations were mainly due to different analytical appliances, software

versions used, the exact amounts of TMU, BCR-656 and lock substance, and differences in

calculations as outlined below (correction of water content or not, etc.). Methodical details

such as NMR spectrometer type, software and QC measures are listed in table 1 (Annex E).

Each laboratory received three units of IRMM-425 randomly selected from the batch.

Laboratories were asked to analyse the samples in duplicate under intermediate precision

conditions, i.e. 2 days à 3 measurements, whereby replicates of a given sample had to be

analysed on different days (day 1: sample 1A, 2A, 3A; day 2: samples 1B, 2B, 3B).

Appropriate performance of instrument and method had to be indicated by providing data on

QC samples before and after the series. Laboratories were asked to provide all raw data and

calculation of the D/H values, a representative spectrum and methodical details. Data were scrutinized at IRMM and eventually re-calculated, whereby the following

considerations applied:

• The value for one sample (replicate), being the mean of 10 acquired spectra, can be

calculated in two ways: either the average of the peak heights (areas) for each signal is

taken, and the respective values are then used in the formula to calculate the D/H value

(approach 1), or the obtained ethanol and TMU peak heights (areas) in each spectra are

taken and used in the formula, and then the average of the 10 calculated values is taken

as the result for one sample (approach 2); the differences were found to be very minor in

all cases (< 0.01 ppm). Approach 2 was used to re-calculate data (when necessary) at

IRMM.

22

• Minor rounding differences (< 0.02 ppm) were sometimes observed between the values

calculated and submitted by the laboratory, and re-calculations of the provided raw data

performed at IRMM.

Laboratory-specific remarks:

Lab 3: The result of one spectrum in each of two of the samples was indicated to be an

outlier (in-house laboratory statistical analysis). However, those values were retained and

used for the calculation of the sample values at IRMM.

Lab 5: Due to a lack of sufficient amount of BCR-656, only 1.7 mL of ethanol could be used

for preparation of the last sample (usual amount is ca. 3.2 mL per sample); however, the

result was retained in the laboratory data set.

Lab 6: In one sample, considerably worse repeatability was obtained among peak heights

within the acquired 10 spectra. The laboratory reported no outliers applying their in-house

statistics, and the overall mean of the sample complied well with mean values of the other

samples. The value was thus retained. Furthermore, this laboratory was the only one where

a considerable difference for the TMU value was obtained when using either the D/HI value

of BCR-656 or its D/HII value. No reasonable explanation could be given for this finding.

However, as explained earlier, only the values calculated using the D/HI value of BCR-656

are taken into account for establishing the certified value.

Lab 7: One of the ampoules provided was also checked for the water content, which was

found to be < 0.11 % m/m. Due to a lack of sufficient amount of BCR-656, only 1.9 mL of

ethanol could be used for preparation of the last sample (usual amount is ca. 3.2 mL per

sample); however, the result was retained in the laboratory data set.

Lab 8: One of the ampoules provided was also checked for the water content, which was

found to be < 0.11 % m/m. Due to a lack of sufficient amount of BCR-656, a laboratory in-

house ethanol (established D/HI value of 103.00 ppm) was used instead for the preparation

of the last sample; the result was retained in the laboratory data set.

Lab 9: One of the ampoules provided was also checked for the water content, which was

found to be < 0.11 % m/m.

Lab 10: The D/H value for TMU was calculated by correcting for residual water (0.16 % m/m

as determined by Karl Fischer titration at the laboratory); furthermore, rounded molecular

masses for TMU and ethanol of 116 and 46 g/mol were used, respectively. Therefore, the

TMU D/H values were recalculated at IRMM, using the coefficients in 2676/90 and leaving

out the correction for the residual water.

23

Lab 11: The laboratory re-examined the alcoholic grade of the 2 bottles of BCR-656 provided

and reported values of 94.38 and 94.45 (% mass, w/w), respectively, which were then used

in the calculation of the TMU D/H value; therefore, the TMU D/H values were re-calculated at

IRMM. The laboratory repeated the analysis of the 3rd ampoule due to instrumental problems

in the first series for this sample. Analysis was done on 2 different days.

Finally, all data sets were subjected to visual and statistical analysis, employing SoftCRM

[11]. These analyses included Dixon test, Nalimov t-test and Grubbs test to detect possible

outliers, Cochran and Bartlett test to verify homogeneity of variances, and skewness and

kurtosis test to check for normal distribution of laboratory means.

Outlier testing revealed one outlier (laboratory 11) with the Nalimov t-test, whereas the other

two tests did not indicate any outliers. After data inspection (laboratory data including the QC

samples), it was decided to retain this value for the calculation of the overall mean. Both

Cochran and Barlett test indicate the homogeneity of variances, and skewness and kurtosis

test revealed normal distribution of laboratory mean values. In conclusion, no data was

rejected, and uchar was calculated as

nsuchar =

s being the standard deviation of the laboratory means, and n the number of accepted data sets (laboratories).

24

Figure 4. Mean of the laboratory means and its standard deviation, and means from all accepted data sets ± standard deviations (n=6) from the characterisation of IRMM-425

Table 7. Mean of the means of the accepted data sets of results, standard uncertainty from characterisation (uchar) and relative standard uncertainty from characterisation (uchar%)

Property Mean of means 1)

Accepted data sets

(p)

uchar1)

uchar [%] 2)

D/H ratio 142.147 x 10-6 11 0.092 x 10-6 0.065 1) Mean value is dimensionless; commonly it is expressed in parts per million (ppm) 2) Relative uchar (in percent of the certified value)

The overall mean value finally was corrected for the purity of the material (99.81 % m/m,

page 9), thus yielding a certified value of 141.877 x 10-6 (141.877 ppm).

8.2 Certified value and uncertainty budget

Uncertainties of the certified value of an individual unit (UCRM) were calculated using the

formula depicted below. In addition to the usually included contributors for homogeneity,

long-term stability, and characterisation, also the uncertainties carried over from BCR-656

25

(D/HI value, alcoholic strength value) were taken into account. The calculated uncertainty

due to purity, however, was not included due to its negligible contribution to the total

uncertainty.

ba BCRBCRcharltsbbCRM uuuuukU 6562

6562222*

−− ++++⋅=

where UCRM is the expanded uncertainty of the certified value

k is the coverage factor; a factor of 2 is applied to give approximately 95 %

confidence

ubb is the uncertainty associated to the between bottle variation

ults is the uncertainty contribution from the estimation of the long-term stability study

uchar is the uncertainty associated to the characterisation of the material

uBCR656a is the uncertainty of the D/HI value of BCR-656

uBCR-656b is the uncertainty of the Dmt value of BCR-656

Table 8. Certified value and uncertainty budget for D/H ratio of IRMM-425 IRMM-425

u*bb [%]1) 0.060

ults [%]1) 0.179

uchar [%]1) 0.065

uBCR-656 a [%]1) 0.089

uBCR-656 b [%]1) 0.024

upur [%]1) negligible

UCRM (k=2) [%]1) 0.439

Average 2) 141.877 x 10-6

UCRM (k=2) 2) 0.623 x 10-6

1) The relative uncertainties are listed (percent of the certified value)

2) Values are dimensionless; commonly they are expressed in parts per million (ppm)

26

9 Metrological Traceability

The certified value is traceable to V-SMOV by strictly adhering to the Community reference

method (SNIF-NMR) as defined in Commission Regulation 2676/90 [1]. For V-SMOW, a

value of 155.76 ppm was used (Tellus (1970) 22: 712-715). The certified value is commonly

expressed in parts per million (ppm).

10 Instructions for use

10.1 Intended use

IRMM-425 constitutes a master batch to which future secondary batches will be linked. The

material is not for sale, and shall be used for the value assignment of the D/H ratio in future

secondary TMU batches.

10.2 Storage

The ampoules shall be stored at a temperature of -20 ºC.

10.3 Minimum sample intake

The minimum samples intake is 1.3 g of TMU when using 10 mm diameter NMR tubes

(Commission Regulation 2676/90, [1]).

10.4 Use of the certified value

The material shall be used as internal standard for the determination of the D/H ratios of wine

ethanol according to Commission Regulation 2676/90.

10.5 Safety precautions

The following health and safety clauses apply:

R 20/22 Harmful by inhalation, and if swallowed R 63 Possible risk of harm to the unborn child

27

11 Acknowledgments

The authors would like to thank M. Lees (Eurofins) and C. Guillou (JRC/IHCP/BEVABS) for

fruitful discussions on a suitable strategy for the certification of this master batch, T. Linsinger

and G. Roebben (IRMM) for reviewing certification report and certificate, and P. Gowik

(Federal Office of Consumer Protection and Food Safety, DE), E. Nordkvist (National

Veterinary Institute, SVA, SE), and J. de Boer (Institute for Environmental Studies, Vrije

Universiteit Amsterdam, NL) as Certification Advisory Panel members, for their constructive

comments.

12 Annexes

Annex A: Results of the homogeneity study

Annex B: Results of the short-term stability study

Annex C: Results of the long-term stability study

Annex D: Characterisation – method-related information

Annex E: Results of the Characterisation study

28

13 References [1] Commission Regulation (EEC) No. 2676/90 of 17 September 1990 determining

Community methods for the analysis of wine. Off. J. Eur. Communities 1990; L272: 1.

[2] Martin, G.J., Martin, M.L. (1981) Deuterium labelling at the natural abundance level as

studied by high field quantitative 2H NMR. Tetrahedron Lett. 22, 3525.

[3] Commission Regulation (EC) No. 2729/2000 of 14 December 2000 laying down detailed

implementing rules on controls in the wine sector. Off. J. Eur. Communities 2000; L316:

16.

[4] Guillou, C., Martin G.J. (1993) Characterisation of reference tetramethylurea (TMU) for

the determination of H/D ratio in alcohols by SNIF-NMR. EUR 14396 EN.

[5] Breas, O., Thomas, F., Zeleny, R., Calderone, G., Jamin, E., Guillou, C. (2007)

Performance evaluation of elemental analysis/isotope ratio mass spectrometry methods

for the determination of the D/H ratio in tetramethylurea and other compounds – results of

a laboratory inter-comparison. Rapid Commun. Mass Spectrom. 21, 1555.

[6] Martin, G.J., Naulet N. (1988) Precision, accuracy, and referencing of isotope ratios

determined by nuclear magnetic resonance. Fresenius Z. Anal. Chem. 332, 648.

[7] Guillou, C., Remaud, G., Lees, M. (2001) The certification of the carbon-13 and

deuterium isotopic ratio content and the alcoholic grade of ethanol from wine (BCR-656,

96 % vol), the carbon-13 isotopic ratio content of sugar (BCR-657), the oxygen-18

isotopic ratio content of water from wine (BCR-658, 7 % vol, and BCR-659, 12 % vol),

and the carbon-13 and deuterium isotopic ratio content and the alcoholic grade of ethanol

from wine (BCR-660, 12 % vol). EUR 20064 EN.

[8] Bau, A., Oostra, A., de Vos, P., Charoud-Got, J., Zeleny, R., Emteborg, H. private

communication (RM unit internal report "Filling of tetramethylurea (TMU) in ampoules.

For a master batch to establish the D/H ratio by NMR", GE/RM Unit/15/2005/October27,

IRMM 2005).

[9] Martin, G.J., Trierweiler, M., Ristow, R., Hermann, A., Belliardo, J.-J. (1994) The

certification of the three reference ethanols for SNIF-NMR: BCR certified reference

materials. EUR 15347 EN.

[10] Cremonini, M.A., Tacconi, D., Clementi, V., Luchinat, C. (1998) Accurate Determination

of Deuterium/Hydrogen Ratios in Natural Organic Compounds through a Nuclear

Magnetic Resonance Time-Domain Reference Convolution Method: Application to

Ethanol from Three Botanical Sources and Critical Analysis of Systematic Inaccuracies in

Previous Methods. J. Agric. Food Chem. 46, 3943.

[11] SoftCRM ver. 2.0.10 developed by NHRF and JRC/IRMM. Customized version of

SoftCRM http://www.eie.gr/iopc/softcrm/

29

[12] van der Veen, A.M.H., Linsinger, T.P., Pauwels, J. (2001) Uncertainty calculations in

the certification of reference materials, 2. Homogeneity study. Accred. Qual. Assur. 6, 26.

[13] Lamberty, A., Schimmel, H., Pauwels, J. (1998) The study of the stability of reference

materials by isochronous measurements, Fres. J. Anal. Chem., 360, 359.

[14] Linsinger, T., Pauwels, J., Lamberty, A., Schimmel, H., van der Veen, A.M.H.,

Siekmann, L. (2001) Estimating the uncertainty of stability for matrix CRMs. Fres. J. Anal.

Chem. 370, 183.

30

Annex A - Homogeneity data Results of the homogeneity study of IRMM-425. Data was calculated according to formula on page 12 [1], taking the (D/H)I value from BCR-656 into account.

Ampoule number Replicate 1 (ppm) Replicate 2 (ppm) Average (ppm) Series 1, (D/H)I signal taken

50 141.94 141.94 141.94 115 141.64 141.95 141.80 166 141.74 142.03 141.89 258 141.75 141.72 141.74 300 141.93 141.82 141.88 354 141.84 141.86 141.85 438 141.98 141.66 141.82 521 141.92 141.52 141.72

Figure A1. Homogeneity of IRMM-425. The x axis depicts the sample numbers (filling sequence). The indicated points are mean values of duplicate measurements.

31

Figure A2. Homogeneity of IRMM-425. The x axis depicts the order of measurements (analytical sequence).

32

Annex B – Short-term stability Results of the short-term stability study of IRMM-425. Data was calculated according to formula on page 12 [1], taking the (D/H)I value from BCR-656 into account.

Time (weeks) D/HTMU (in ppm) 18°C 60°C 0 141.71 141.71 0 141.48 141.48 0 141.59 141.59 0 141.55 141.55 0 140.92 140.92 0 141.45 141.45 1 141.58 140.97 1 141.69 141.46 1 141.49 141.53 1 141.13 141.56 1 141.51 141.88 1 141.42 140.84 2 141.00 141.41 2 141.35 141.61 2 140.61 141.07 2 141.59 141.07 2 141.26 140.91 2 141.52 140.86 4 141.64 141.65 4 141.13 141.07 4 141.46 142.10 4 141.44 141.69 4 141.27 141.37 4 141.61 141.24

33

Figure B1. Short-term stability of IRMM-425 at 18 ºC.

Figure B2. Short-term stability of IRMM-425 at 60 ºC.

34

Annex C – Long-term stability Results of the long-term stability study of IRMM-425 at -20 ºC. Data was calculated according to formula on page 12 [1], taking the (D/H)I value from BCR-656 into account.

Time (months) D/HTMU (in ppm) 0 141.55 0 141.47 0 141.34 0 141.14 0 141.34 0 141.68 0 141.52 0 141.82 0 141.52 0 141.61 0 141.26 0 141.24 12 141.37 12 141.77 12 141.71 12 141.44 12 141.09 12 141.18 12 141.23 12 141.82 12 141.57 12 141.55 12 141.78 12 141.45 18 141.61 18 141.66 18 141.29 18 141.55 18 141.58 18 141.40 18 141.41 18 141.57 18 141.56 18 141.41 18 141.58 18 141.51

35

Figure C1. Long-term stability of IRMM-425 at -20 ºC. ults is estimated for a shelf life of 5 years using the formula on page 17 and the measurement data on page 32.

36

Annex D – Characterisation – method-related information Laboratory

code Signal

property used for calculation

NMR appliance and software

QC measures (instrument

verification, QC sample(s) used)

1 Height Bruker Avance 400, Topspin 1.3

BCR-656, samples of latest TMU batch

included in sequence

2 Height Bruker Avance 400, Topspin 1.3

BCR-123, in-house alcohol

3 Height Bruker Avance 400, X-Win NMR

BCR-123, TMU calibrated in-house

4 Height Bruker Avance 400, Icon NMR

BCR-123, in-house wine ethanol sample

(95 vol%) 5 Height Varian Unity Inova 300,

VNMRJ rev. 1.1D BCR-123

6 Area Bruker ARX 500, X-Win NMR 2.6

BCR-123

7 Height Bruker AM 500, Aspect X32

BCR-123

8 Height Bruker AMX 500, X-Win NMR 2.6

BCR-123

9 Height Bruker DPX 400, Topspin 1.3.6

BCR-123

10 Height Bruker Avance 400, X-Win NMR 3.1

BCR-123

11 Height Bruker DXR 500, X-Win NMR 2.3

BCR-123

37

Annex E – Characterisation data Characterisation data of IRMM-425. Data (ratio of deuterium to hydrogen) is expressed in ppm and has been acquired from TMU/BCR-656 mixtures strictly adhering to 2676/90 [1]. Values were calculated using the D/HI value of BCR-656 and the formula depicted on page 12. All 6 individual measurements (2 days, 3 samples in duplicate) per laboratory are presented. Data were eventually re-calculated at IRMM performing a harmonised way of calculation (see page 22).

Laboratory code

Day 1 Day 2

Sample 1 Sample 2 Sample 3 Sample 1 Sample 2 Sample 3

1 141.42 141.26 141.91 141.93 141.83 142.02 2 142.48 142.06 142.28 141.82 142.10 142.19 3 142.36 142.15 142.36 142.40 141.84 142.24 4 141.72 142.14 142.26 142.00 141.74 142.30 5 142.08 142.30 142.19 141.57 141.59 141.72 6 142.34 142.39 141.86 142.27 142.93 142.41 7 142.61 142.52 142.63 142.53 142.86 142.56 8 142.31 142.14 141.57 142.23 142.11 141.93 9 142.09 141.92 141.97 141.97 142.12 142.03

10 141.85 141.82 142.41 141.47 141.56 142.04 11 143.34 143.58 143.12 142.22 142.29 141.69

European Commission EUR 23031 EN – Joint Research Centre – Institute for Reference Materials and Measurements Title: Certification of the Deuterium-to-Hydrogen (D/H) ratio in a 1,1,3,3 – tetramethylurea master batch, IRMM-425 Author(s): R. Zeleny, H. Emteborg, F. Ulberth Luxembourg: Office for Official Publications of the European Communities 2007 – 37 pp. – 21.0 x 29.7 cm EUR – Scientific and Technical Research series – ISSN 1018-5593 ISBN 978-92-79-07550-6 Abstract This report describes the production of a tetramethylurea reference material (IRMM-425), certified for its

deuterium-to-hydrogen (D/H) ratio. The material is to be used as an internal standard in site-specific natural

isotope fractionation – nuclear magnetic resonance (SNIF-NMR) spectroscopy measurements for determining

the D/H ratios of ethanol distilled from wines, an important measure in wine authenticity testing (Commission

Regulation 2676/90, [1]).

Commercially obtained TMU with a sufficiently high D/H ratio (>120 x 10-6) was purified by removing most of the

residual water and filled into amber glass ampoules in 7.8 mL portions. Homogeneity was tested and no

heterogeneity observed. Stability studies indicated no material degradation for 4 weeks at 60 ºC and for 18

months at –20 ºC. Batch characterisation was accomplished in an inter-laboratory comparison using the SNIF-

NMR technique exclusively.

The certified value was obtained as the unweighted mean of the laboratory means of the accepted sets of

results. The expanded uncertainty associated (k = 2) includes contributions from (potential) heterogeneity,

potential instability, characterisation, as well as the carried-over uncertainty from BCR-656 (96% wine ethanol)

that was used as the internal standard in the measurements. Due to its negligible contribution, the uncertainty

associated with the purity of the material did not need to be included.

The certified value and its associated uncertainty are:

Amount-of-substance fraction

Certified value 1,2,4) Uncertainty 3)

Deuterium-to-hydrogen (D/H) ratio 141.9 x 10-6 0.7 x 10-6 1) Traceable to V-SMOV (Vienna Standard Mean Ocean Water) by strictly adhering to the Community

reference method (SNIF-NMR, site-specific natural isotope fractionation – nuclear magnetic resonance spectroscopy) as defined in Annex 8, Commission Regulation 2676/90. For V-SMOW, a value of 155.76 ppm was used (Tellus (1970) 22: 712-715).

2) This value was obtained by an inter-laboratory comparison (11 laboratories) employing SNIF-NMR at 46.1 (300), 61.4 (400), and 76.7 (500) MHz for 2H (1H). The value has been corrected for the material impurities (including water).

3) Expanded uncertainty with a coverage factor of k = 2, according to the Guide to the Expression of Uncertainty in Measurement, corresponding to a level of confidence of about 95 %.

4) The certified value has a measurement unit of one; it is commonly expressed in parts per million (ppm).

This CRM constitutes a master batch and will be used for the certification of secondary batches; the material is

not for sale.

The mission of the JRC is to provide customer-driven scientific and technical support for the conception, development, implementation and monitoring of EU policies. As a service of the European Commission, the JRC functions as a reference centre of science and technology for the Union. Close to the policy-making process, it serves the common interest of the Member States, while being independent of special interests, whether private or national.

LA

-NA

-23031-EN-C